A Comprehensive Review of the Utilisation of Artificial Intelligence in the Maintenance of Railway Infrastructure

The evolution of Artificial Intelligence (AI) technologies has seen an excessive utility across several industrial domains including railways. In general, railway infrastructure is composed of various components like track, overhead electrical systems, signaling system and communication system. The name of these components varies with different countries; for instance, Swedish inference for the railway infrastructure components is named as Bana, El, Signal, Tele (BEST) corresponding to track, electricals, signaling and telecom. The purpose of this paper is to provide a comprehensive review on the AI technologies utilized in the maintenance of railway infrastructure. Thereby highlighting the significance of AI in facilitating the execution of maintenance activities carried out in railway infrastructure. In this study, a total of 99 scientific papers were reviewed that were published between January 2019 and April 2024. The selected papers were reviewed based on the AI techniques used in the maintenance of railway infrastructure and were further categorized into major AI technologies within components in the railway infrastructure. The analysis based on the literature survey states that most of the operation and maintenance activities revolved around the detection of track and catenary defects. Whilst, only limited or no research was found related to AI in the fields of signaling and telecom. However, the adoption and application of AI in maintenance of railway infrastructure are still at the infant stages. A large scope persists in developing a combined application of AI and multitude of techniques such as feature fusion, point cloud data, image analysis etc., that can be adopted for effective decision-making, enhanced optimisation, ease to handle uncertainties and tackle cybersecurity related issues.